KAIA Model Gate icon

KAIA Model Gate

Intercept LLM input and output to record all interactions for AI Act compliance

KAIA Model Gate

AI-generated

Overview

The KAIA Model Gate node intercepts input and output calls of a connected language model on the n8n canvas to record all interactions. It is designed to ensure AI Act compliance by capturing and sending detailed signals about the language model's input prompts and output responses to an external Kaia API service. This node is useful in scenarios where organizations need to monitor, log, and audit AI language model usage for regulatory or compliance purposes.

Use Case Examples

  1. Intercepting and logging all prompts and responses of a language model used in an automation workflow to maintain an audit trail.
  2. Monitoring token usage and latency of language model calls to optimize costs and performance.
  3. Ensuring compliance with AI regulations by recording detailed interaction data with language models.

Properties

Name Meaning
Connect a language model on the canvas to intercept its calls This notice informs the user that the node requires a connected LangChain language model input to function properly.

Output

JSON

  • response - The intercepted language model's response object, including text output and token usage details.

Dependencies

  • Requires Kaia API credentials (base URL and token) to send intercepted data to the Kaia compliance service.

Troubleshooting

  • Error 'Kaia API credentials are required' indicates missing or invalid API credentials; ensure credentials are configured correctly.
  • Error 'No model connected' means the node is not connected to a language model input; connect a compatible LangChain model to the node's input.
  • Network errors when sending signals to the Kaia API may occur; verify network connectivity and API endpoint availability.

Links

  • Kaia AI - Official website for the Kaia AI compliance service used by this node.

Discussion